Dear researcher

please check this attachments.

http://www.ie.boun.edu.tr/~taner/publications/papers/membership.pdf

http://dopti.com/~dombi/publications/1988-J.Dombi---Membership_function_as_an_evaluation.pdf

Use GA or PSO techniques to optimize your fuzzy intervals. I advise you to read relevant papers dealing with such problems in other fields such as in " Power System Control"

dear researcher

this attachments will also be helpful for you.additionaly you can refer to my publications where i have used the rules and appropriate intervals.

Regards

http://herkules.oulu.fi/isbn9514264355/isbn9514264355.pdf

Some times, it makes difficult to define the membership function. so you can first define the fuzzy rules using fuzzy inference system. after that u can define the membership function 

A fuzzy membership function is a range to represent a condition which could be as simple as the fever is 'low', 'medium' or high or the pain is 'severe' , 'medium' etc...

You need to read several papers on the problem which you wish to address in order to check if you can get this range from the available literature. If not, you need to 

consult several medical Experts to identify and optimize this range and develop an Expert system. Fuzzy rules can be framed by consulting several Experts. This can be done in case when the information is linguistic and/or ambiguous.  You can determine the appropriate intervals in fuzzy membership functions based on the range you acquire and develop membership functions which represent this range/intervals. At an initial step you can use the 'fuzzy' tool in MATLAB and develop a Fuzzy Inference System.